www.cmm.bristol.ac.uk/forum

Dear all,
I have to fit multilevel discrete time hazard model to examine correlates of third birth. For the purpose, I have converted my data into person period format as required for the discrete time survival analysis. Now, I am trying to fit a two level model using runmlwin. In the mdoel I have specified a women to be at level 2 and each person-period at level1 one.
When I run the model with this specification, the women level variance is coming to be zero.
I am not sure if this specification is correct. I also have to explore community level effetcs at level three.
Please help me in this.
The stata syntax I am using is given below:

In the syntax, id1 is women id and id2 is id for each period.

#delimit ;
runmlwin d2 cons t1-t11 ,
level2(id1:cons ) level1(id2)
discrete(distribution(binomial) link(logit) denominator(cons) pql2)
mlwinsettings(size(536000) levels(5) columns(1500) variables(300) optimat) nopaus;
#delimit cr

Dear rahulvbb,

The specification looks sensible enough to me.

Perhaps you have run into estimation models, first try fitting the model by MQL1 which tends to have less convergence difficulties.

You can add community as a third-level to the model in the usual way

level3(community: cons)

however, there is no point doing this while you have having difficulties with this simpler model.

You might want to consider first using runmlwin to replicate some of the MLwiN point-and-click discrete time hazard models found on the CMM website before trying to get these models working on your own data.

http://www.bristol.ac.uk/cmm/research/m ... /#workshop

Best wishes

George